Deciphering how the brain processes sensory information and makes behavioral decisions ultimately requires methods to record simultaneously from all the neurons in a local circuit. Of techniques used to measure neuronal activity, optical microscopy stands out as one of the few tools with the spatial resolution needed for dense ensemble recordings in thick tissue. Recently, significant progress has been made in developing probes and instrumentation for measuring neuronal activity using fluorescence. For studies of neural circuits, however, this strategy is fundamentally limited by the phototoxicity of the fluorophore: long-term, high-speed imaging needed to study circuits delivers a light dose that damages (and ultimately destroys) fluorescently-labeled cells. To overcome this problem, I propose to develop new approaches to measure spiking activity without fluorescence. The methods we will develop will perform at speeds sufficient to image large three-dimensional volumes of intact tissue thousands of times per second, sensitive enough to record each action potential, and sufficiently non-damaging to record from the same set of neurons for periods of hours. This technical advance will provide an unprecedented look at how neuronal activity generates the central computational functions of the brain.